Chapter 9. Conclusions and Implications

Carol L. Council, M.S.P.H.
Jeremy W. Bray, Ph.D.

Introduction

Over the past two decades, the behavioral health care delivery system in the United States has experienced rapid change in terms of clients served, as well as organizational and financial structures. New patterns of substance use, the changing demographics of affected populations, the availability of pharmacotherapies, and the growing presence of managed care have challenged the system and its study. These changes are the result of competing demands to contain costs, maintain the quality of care, make care available to all who need treatment, and focus resources on those forms of care that offer the best hope of successful outcomes. Building upon such disciplines as epidemiology, biostatistics, and health economics, as well as the impressive body of health services research findings, the field of behavioral health research has been called on increasingly to analyze these changes and their effects.

The chapters in this compendium extend our knowledge by providing information on factors that may facilitate or inhibit the delivery of behavioral health services for those persons with substance use or mental disorders. Valuable new information is provided on accessibility, utilization, quality, cost, cost-effectiveness, and outcomes. The findings describe the relationship between demographic characteristics and how, when, where, and if a person will seek care, as well as what types of care are chosen and what happens during the delivery of care. Finally, the findings presented in this compendium also reveal potential biases that may occur when using many of the large datasets currently available for conducting research in the behavioral health sciences.

This chapter explores the implications of these findings in key areas for policymakers, the treatment community, and researchers and suggests directions for future research.

Access, Financing, and Costs of Substance Abuse Treatment and Mental Health Services

Several of the chapters in this compendium explored the issue of access to care and underscored the role of managed care in bringing financing and cost issues to the forefront as a potential barrier to client access to needed services. Dr. Woodward in his Chapter 2 literature review, identified access as a critical first step to getting clients needed treatment. Although it is generally accepted that people with substance use and mental disorders may be affected severely by access problems, individuals with more serious behavioral illnesses may encounter additional barriers to access because successful treatment may be more expensive, because redundant and bureaucratic procedures may pose insurmountable obstacles, because needed services may not be available, and because the private-payer managed care system may benefit financially from placing barriers to treatment on such people. On the positive side, reports indicate that a few State-sponsored programs have been able to increase access for high-risk populations.

In Chapter 4, Dr. Duffy and colleagues examined the factors influencing admission to treatment and reported that in most States examined, the criminal justice system was the most common source of treatment referral for adult males, with alcohol as the primary substance of abuse. They also showed that referral by an alcohol or drug treatment provider generally increased the likelihood of inpatient admission. Clients who were employed were less likely to have an inpatient admission, and clients who were expected to pay for their own care had a lower likelihood of entering inpatient treatment. In many of the States examined, co-occurring mental disorders did not consistently increase the odds of inpatient admission as would be expected given the accepted clinical placement criteria at the time.

In Chapter 7, Dr. Duffy and colleagues estimated cost functions for substance abuse treatment programs as a function of size. Findings indicated that the average cost per admission declined as facilities became larger, which suggests that larger facilities may be able to provide care at a lower price than smaller facilities. These economies may suggest policies that encourage the use of large facilities.

Implications for Access for High-Risk Populations

The research presented suggests that the effects of benefit and financing factors on access to care may be different for behavioral health care clients than for medical care clients. Policymakers may wish to consider this when weighing the adoption of financing and delivery mechanisms used to finance care. Rather than preventing unnecessary use of resources (i.e., moral hazard), aggressive cost sharing and case management policies may deter medically necessary and potentially cost-beneficial substance abuse treatment and mental health services. In fact, such policies may force providers to avoid providing care to expensive high-risk populations, such as homeless persons or those with co-occurring disorders, by creating administrative barriers that effectively exclude those with the greatest need.

Variability in client level of need and service provision across programs affects client outcomes. Reports of effective programs providing limited treatment services should be reviewed by policymakers with care to be certain that clients served in those programs have levels of need similar to those in other programs. Although it may be tempting to provide fewer costlier services, treatment research suggests that optimal length of stay should be determined on a case-by-case basis, using all available clinical information rather than dictated by a universal policy. Although a shorter length of stay may contain short-term costs, it may not contain long-term costs; thus, clients may end up costing both society and the health care system much more in the long run. At a minimum, treatment-funding systems may be required to provide costlier services to populations on SSI or SSDI because these populations, in general, manifest more severe problems related to their substance use. As discussed in Chapter 7, facilities with a greater proportion of clients who received supplemental security income (SSI) or social security disability insurance (SSDI) had higher costs.

Implications for Service Delivery

Several of the large national evaluations discussed earlier in this volume reported shifts from inpatient to outpatient care and suggested that the quantity and intensity of services to clients are decreasing. Further, there is some suggestion that programs are being eliminated or are merging into organized networks of behavioral care that are able to provide a wider continuum of services. These shifts may be very beneficial in terms of providing access to a larger pool of clients in need of treatment. Indeed, Dr. Duffy and colleagues in Chapter 7 reported that the average cost per admission declined as facilities became larger, suggesting that larger facilities may be able to provide care at a lower price than smaller facilities. On the other hand, pressure to reduce costs may result in substantial reductions in the quantity and quality of treatment services, especially in those systems that manage resources through a defined level of service benefit.

Although such approaches may help to extend the reach of treatment services to more individuals, in-depth study is needed to explore whether high-risk clients with more severe substance use disorder-related problems are getting the intensity of care they need. Clearly, better information about both severity and actual service parameters provided to all treatment clients is necessary to understand the impact of these changes on treatment access, retention, and outcomes. In addition, research is needed to determine whether client outcomes vary with facility size (particularly outcomes for high-risk populations). Research also needs to determine whether higher costs are unavoidable with certain populations, or whether alternative treatment technologies or administrative arrangements could reduce costs without negatively affecting outcomes.

Implications for Treatment Outcomes

At the provider level, if provider organizations grow larger and more centralized to capture economies of scale, what impact will this have on treatment access in rural areas? Further, what impact will streamlining of providers' treatment regimes have on their ability and willingness to try new approaches for treating clients and/or participate in research? Just as a host of research in the late 1980s and early 1990s examined the effect of hospital mergers and acquisitions on medical outcomes (Finklestein, et al., in press; Hoerger, 1991), behavioral health services researchers might now explore the effects of treatment program scale and scope on substance use and mental disorder outcomes, such as treatment retention, relapse rates, and social functioning.

Implications for Public- and Private-Sector Care

The impact of changes in private-sector financing strategies on the public sector needs to be better understood. When the Federal Government made the decision to shift funding of behavioral health care to the States, it was implied that States would provide services for the neediest and most fragile portions of their populations. During the past decade, private health insurance coverage for behavioral health problems declined at the same time that an increasing number of Americans had no health insurance at all. This has placed an added burden on public behavioral health care systems. Thus, a larger and more diverse array of clients may present to public treatment programs. Moreover, as Dr. Woodward discussed in Chapter 2, treatment facilities that depend on public funding may not have sufficient capacity to provide services to all individuals demanding treatment services, let alone those who have not yet sought service. More information is needed regarding the level of funding required to provide services to uninsured persons and ways of making such funding cost-effective. Clearly, in order to study this issue, we need detailed information on the types of services provided to specific types of clients in both the public and private sectors, and we need information on clients who have exhausted their private-payer benefit for substance abuse treatment and mental health services.

Community Hospitalizations of Those with Substance Use and/or Mental Disorders

Dr. Duffy in Chapter 3 discussed in detail the impact of access and financing with regard to community hospitalizations of those with behavioral health problems. As Dr. Duffy reported, managed care and behavioral health care carve-outs usually result in reductions in inpatient mental health services and inpatient substance abuse treatment. She reported that, contrary to expectations, the discharge rate for individuals with substance use and mental disorders in community hospitals increased over the period from 1990 to 1995, with increases being greater for patients with co-occurring disorders and those aged 35 to 45. Even though the patients who were admitted had more complex disorders, their lengths of stay decreased. Patients with substance use and mental disorder diagnoses were more likely to receive uncompensated care or to have Medicaid coverage than other community hospital patients.

Implications for the Public Sector

The Federal Government's role in paying for the care of substance use and mental disorder patients in community hospitals increased in the early 1990s, with Medicare and Medicaid paying for the treatment of more than half of discharges with such diagnoses. If the Federal Government is paying for a large portion of this care, it may be in a position to develop incentives to ensure that high-risk patients receive adequate treatment support when returning to the community. At a minimum, research is needed to clarify the impact of system changes in substance abuse treatment and mental health services on access to care for the most fragile populations and perhaps to explore the possible need for less expensive residential care options for those patients requiring longer periods of care.

Implications for Community Hospitals and Providers

Many inpatient treatment facilities, such as mental health hospitals and residential substance abuse treatment facilities, closed during the 1990s. The increased number of substance use and mental disorder admissions to community hospitals during the early 1990s suggests that patients with more severe disorders formerly served by other types of inpatient treatment facilities may now be presenting to community hospitals. In turn, community hospitals with resources depleted by reductions in standard lengths of stay for traditional patients may have provided lifesaving services, such as detoxification and stabilization to substance use and mental disorder patients and discharged them back to the community as soon as possible. Community hospitals may not be in a position to provide longer-term residential treatment to such patients. To enhance the likelihood that they will not be readmitted, community hospitals might consider developing strong links with providers of community-based substance abuse treatment and mental health services. This might ensure that high-risk patients transitioning back to the community are engaged in appropriate aftercare programs.

The use of community hospitals as a substitute for State-supported detoxification services and patient stabilization may be financially beneficial if realistic procedures are in place for shifting the management of care at discharge to community-based providers. At a minimum, to determine the level of case management support that such patients need when they return to the community, providers need information on the level of functioning of discharged community hospital patients, as well as the specific services the patients received while in the community hospital. In addition, it would be beneficial to understand whether the patients served were those who were unable to access needed care earlier in a community setting.

Treatment Retention

In Chapter 6, Dr. Ashley and colleagues examined the role of women-only treatment programs in retaining women in substance abuse treatment. They found that substance abuse treatment for women at facilities offering child care services and treatment at women-only facilities were associated with longer stays. However, lower educational levels among female clients were associated with shorter stays.

Implications for Treatment of Specific Populations

Based on these findings, policymakers may want to develop financing systems to encourage the development of gender-appropriate treatment programs. Tailoring treatment programs to meet special needs of certain population groups not only makes treatment more attractive to those who need it, but it also helps to address many of the logistical and pragmatic barriers to treatment faced by special populations.

It is understandable that behavioral health sciences research has focused largely on the neediest and most fragile portions of the substance use and mental disorder population. Given the prevalence of substance use and mental disorders across all population groups, it may be appropriate to focus on the impact of shifts in financing and access on access for other population groups as well. In Chapters 4 and 7, Dr. Duffy and colleagues reported that employed persons are less apt to receive inpatient treatment. Furthermore, lengths of stay in treatment are shorter for employed women (Chapter 6), suggesting that other segments of the population may be affected by managed care practices. Although it may be hypothesized that employed persons in treatment may have less severe disorders, more information about these groups is needed, including what factors influence treatment-seeking behaviors, their level of functioning, and their support environment. More information is needed about the stigma of having a behavioral health disorder and the extent to which it prevents people from seeking needed treatment or from completing their treatment plan.

Treatment Choice

In Chapter 4, Dr. Duffy and colleagues used the 1996 Treatment Episode Data Set (TEDS) to explore factors related to treatment admission for adult males, with alcohol as the primary substance of abuse. They found that greater substance use disorder severity increased the likelihood of inpatient admission, but in many of the States examined, co-occurring mental disorders did not consistently increase the odds of an admission. Also notable was the fact that referral by an alcohol or drug treatment provider generally increased the odds of inpatient admission. This may reflect changes to the treatment system in which treatment providers serve as gatekeepers to higher levels of care.

Further, Dr. Duffy and colleagues in Chapter 5 examined treatment choice in two States from among five types of treatment (standard outpatient, intensive outpatient, long-term residential, short-term residential, and inpatient hospital) instead of the standard two (inpatient and outpatient). They reported that those admitted to standard outpatient treatment appeared to have less severe alcohol disorders and were more likely to be employed than those admitted to any other treatment setting. Furthermore, they reported that analyses that allowed for only two choices, inpatient and outpatient, obscured the relationships between client characteristics and treatment-setting choice.

Implications for States

In general, it appears that the States are following best practice recommendations with regard to client placement. In many of the States examined, however, co-occurring mental disorders did not consistently increase the odds of inpatient admission as might have been expected given current clinical guidelines regarding placement criteria. This may not be a function of States' not adhering to best practices, but rather may be the result of client choice.

We do not understand fully the differences between States' substance use and mental disorder treatment programs, nor how those differences might affect the organization of treatment at the State level (i.e., where the responsibility for substance abuse treatment and mental health services falls in State government), and the effect that the differences have on access to care for the sickest clients as well as on treatment choice. Many State substance abuse treatment programs are part of State mental health departments, others are part of State health departments, and still others are individual, stand-alone State agencies. In addition, some States have developed mechanisms for sharing data on client populations served by several State agencies. Closer examination of the effect that administrative and organizational factors may have on access to care may reveal those factors that may contribute to better access for high-risk populations. Similarly, differences exist in the extent to which State funding resources are utilized to support substance abuse treatment and mental health services. In order to understand access to care, more information is needed about State and local funding levels for substance abuse treatment and mental health services and their impact on access and treatment.

Implications for Characterization of Treatment Options

To understand treatment choices, researchers need to expand their concept of treatment modality beyond outpatient and inpatient in order to recognize better the full range of treatment modalities now available. More client-level service data that include the full range of treatment options utilized are needed by policymakers so they may account for more complex treatment choices when developing cost-sharing and financing policies.

Data Issues in Behavioral Health Care Research

As reported throughout this compendium, much of the research reported upon in the literature relies on analyses of small, selected, nonrepresentative samples, often with very low response rates and many missing observations. At best, this leads to difficulty in comparing studies and synthesizing findings reported in the literature. At worst, it can lead to the generation of misleading, even false, information. Large-scale data collection efforts, including those at the Federal and State levels (some of which have been used in the chapters of this compendium), have made strides in improving the data available to researchers. However, these data pose challenges as well.

Most state-level substance abuse treatment authorities provide treatment episode data to the Federal Government. However, variations exist at the State level with regard to which facilities must report treatment episode data. In Chapter 8, Dr. Duffy looked at selection biases in data reporting systems between States that require the reporting of substance abuse treatment and mental health services for all clients and those that require the reporting of such data only from facilities receiving public funding earmarked for substance abuse treatment. She found variations across States between clients who entered facilities that accepted earmarked funds and those who entered facilities that did not accept earmarked funds. Clients admitted to facilities receiving earmarked funds generally were younger, less likely to be employed, less likely to be married, less likely to have postsecondary education, and less likely to have private insurance pay for their treatment than those entering other facilities.

Dr. Duffy concluded that analyses of data collected by States only from facilities that receive public funds earmarked for substance abuse treatment may be biased and cautioned that selection effects may bias estimates of the impact that client demographics have on the probability of inpatient admission to publicly funded treatment facilities.

The lack of client-level data on the level of need and care provided during an episode of treatment was mentioned by several authors. The costs per unit of service and per episode of care are frequently unavailable. The absence of this information hampers cost-benefit analysis, as well as outcomes research.

Implications for Federal and State Data Systems

Because not all treatment facilities report data, policymakers may have an incomplete understanding of the potential effects of policy changes. States may want to consider the required reporting of client data by all facilities, regardless of their funding status. If this is not politically feasible, the Federal Government may choose to use synthetic estimation to adjust for State reporting differences in federally maintained databases.

Because performance-based programming is a priority at both the national and State levels, many States will need more detailed cost and service information. The use of uniform client assessment procedures, as well as the development of management information systems that will provide client-level treatment episode data, will enhance performance monitoring.

Conclusions

Recent changes in the organizational and financial structures of the behavioral health care delivery system in the United States have had a large effect on the structure of the behavioral health treatment system and on behavioral health service delivery. These changes have not been adequately studied. This compendium has presented new research that helps to fill this gap. It gives policymakers and service providers at the Federal, State, and local levels a better understanding of how these changes are affecting access to needed care, the quality and effectiveness of care, the utilization of services, cost of treatment services, and the outcomes of treatment for people with acute and chronic substance use and mental disorders.

Several overarching conclusions emerge from the findings presented. Much can be learned by studying the findings of general health services research. Both the behavioral health and the general health systems have been greatly influenced by managed care, and not all the influences have been negative. Managed care entities often use the results of research to guide clinical treatment. In the behavioral health area, research indicating that inpatient care did not provide better outcomes for most clients than outpatient care was used to modify clinical guidelines and resulted in the shift to decreased use of inpatient care.

On the other hand, important differences exist between the behavioral health care system and the general health system. The behavioral health care system has in large part not suffered from overutilization of services. In fact, State systems have been criticized because they have not been able to close the treatment gap by engaging more of the population in need. This compendium underscores the need for better information on the impact that changes in the behavioral health care system have had on access to both fragile high-risk populations and persons needing services in the general population.

Fragile populations (such as individuals who are homeless and displaced youths) may need more intense and longer-term support. As reported for the RAND Insurance Experiment (Newhouse & Insurance Experiment Group, 1994), one such fragile population is affected negatively by managed care approaches to treatment. Populations with substance use and mental disorders frequently experience economic, social, and health problems that may detract from their ability to access and remain in treatment. Providing treatment services to fragile populations presents a difficult challenge to behavioral health care treatment providers because they often require more intensive, lengthy treatment.

As in the health services field in general, nationally based, representative datasets are available for additional study. The existence of large national datasets provide policymakers and researchers with an important base from which to study both the populations manifesting substance use disorders and the systems providing services to these populations. Data collection and management systems, including Web-based systems, have been developed that enable the collection of much richer data on a much broader population of clients and providers. Another key implication of this compendium is that to support effective policy, more information is needed on service need, types of services available, the cost of an entire episode of treatment, and State and local differences in treatment policies. In this age of Web-based reporting, increased opportunities exist to provide this information. In addition, improved data analysis techniques permit the synthesis of information from varied sources.

As an additional step to improving the quality of data collected, clear definitions are needed for assessing clients, and for defining the types of services they receive. Two remaining problems affect the analyses of available data. First, the datasets themselves have biases that should be fully understood before drawing conclusions from them. Second, the information in these systems could be substantially improved. In the past, the field has had to depend on small well-funded studies of service providers that may or may not have represented typical service provision. Although many of these studies yielded compelling data on outcomes, these findings were very difficult to use as the basis for policy because they were narrow and potentially nonrepresentative. Improvements in the quality and quantity of service and client information have been made, but more is needed. In particular, guidance is needed on potential approaches to developing more detailed service and cost information.

Managed care for substance abuse treatment and mental health services has shifted treatment from inpatient to outpatient settings and has shifted the financial risk to providers, thus constraining provider treatment options. Although the shift from inpatient to outpatient settings may be appropriate for many clients, the most severely troubled clients may be in danger of being routed to less effective and less cost-effective care. In addition, the shifting of risk to providers may deter the use of more expensive and intensive therapies that are more cost-effective from a societal standpoint but pose a greater financial risk to providers. Similarly, providers may be unwilling to adopt best practices because of the financial costs of so doing. In a truly capitated treatment system with a stable population of plan enrollees, long-term outcomes have greater value. In managed care plans, such as Medicaid, where clients are constantly switching health care insurance plans, the plans have more incentive to manage short-term costs because long-term costs may be borne by another managed care entity. Research is needed to explore the sharing of risk between the public and private sectors. Health economics research is needed to develop a system that enables the sharing of financial risk among all payers while providing financial support for the adoption of best practices.

This compendium provides additional examples of how health services research in the behavioral health care sector can inform policy and have clear implications for researchers, policymakers, providers, and clients. Although more work is needed in critical areas, such as the role of client characteristics in the treatment system, determining the most cost-effective care modalities, and identifying the minimal set of data reporting programs and characteristics necessary, this compendium clearly shows that health services research can and does play a critical role in the formation of effective policy. It is hoped that the research presented here will help to inspire both new and more effective policies and also new research that will continue to improve the lives of individuals with substance use or mental disorders.